Characterization of microscale features and mineral distributions in rock samples can be facilitated non-destructively with imaging analysis. Scanning electron microscopy combined with backscattered electron and energy-dispersive spectroscopy is particularly valuable and can be utilized to identify minerals. Mineral segmentation coupled with quantitative image processing can yield mineral volume fractions and accessibility from these images. Prior estimates of mineral accessibility from images have improved the simulations of mineral reaction rates, but it is unclear how pore connectivity should be accounted for. This is further complex in samples with clay minerals where nanopores in clays need to be considered. Here, the impacts of different approaches to assess pore connectivity on quantification of mineral accessibility are considered for seven sandstone samples with varying composition. Mineral accessibilities are calculated by counting the interfacial pixels between the associated minerals and the adjacent pores from the 2D mineral segmented maps. Three types of accessibilities are considered: the first approach accounts for all the macropore space, the second approach considers only the 2D connected macropores, and the third approach includes the 2D connected porosity considering nanopores in clays. The observed variations in accessibility for most mineral phases are within 1 order of magnitude when nanopore connectivity is considered and thus not anticipated to largely impact the simulated reactivity of samples. However, greater variations were observed for clay minerals, which may impact long-term simulations (years). Larger variations in accessibility were also noted for carbonate minerals, but only some samples contained carbonate phases, and additional data is needed to assess the trends.

The presence of fractures in caprocks can pose increased risks in subsurface energy systems and processes like CO2 sequestration by introducing high-permeability leakage paths. Fracture apertures and permeability can be altered through mineral dissolution and precipitation reactions, but the reactive evolution of fractures is not well understood. In fractures, minerals that are otherwise inaccessible to reactive fluids can become exposed, resulting in mineral reactions unpredicted by bulk formation data. This work seeks to understand the relationship between mineralogy and fracture formation to enhance our understanding of reactive fracture evolution and CO2 leakage potential. Here, the mineral compositions of mechanically induced fracture surfaces in samples of the Mancos and Marcellus shales have been quantified and compared to those of the near-fracture matrices using imaging and bulk X-ray diffraction (XRD) data. In the Mancos shale, the concentrations of clay minerals are enhanced along fracture surfaces with respect to the bulk, and the fracture is most likely to form at kaolinite–kaolinite interfaces. Evaluation of the mineralogical spatial variability through cross-correlation analysis of the surrounding matrix in images of samples cut perpendicular to the fracture shows that clay is 16.7 times more likely to be present than carbonate minerals near the fracture surface. The high correlation persists roughly 200 μm into the surrounding matrix for the Mancos sample and implies that the fracture formed within a defined clay-rich lithofacies.

No abstract

No abstract

X-ray CT imaging provides a 3D view of a sample and is a powerful tool for investigating the internal features of porous rock. Reliable phase segmentation in these images is highly necessary but, like any other digital rock imaging technique, is time-consuming, labor-intensive, and subjective. Combining 3D X-ray CT imaging with machine learning methods that can simultaneously consider several extracted features in addition to color attenuation, is a promising and powerful method for reliable phase segmentation. Machine learning-based phase segmentation of X-ray CT images enables faster data collection and interpretation than traditional methods. This study investigates the performance of several filtering techniques with three machine learning methods and a deep learning method to assess the potential for reliable feature extraction and pixel-level phase segmentation of X-ray CT images. Features were first extracted from images using well-known filters and from the second convolutional layer of the pre-trained VGG16 architecture. Then, K-means clustering, Random Forest, and Feed Forward Artificial Neural Network methods, as well as the modified U-Net model, were applied to the extracted input features. The models’ performances were then compared and contrasted to determine the influence of the machine learning method and input features on reliable phase segmentation. The results showed considering more dimensionality has promising results and all classification algorithms result in high accuracy ranging from 0.87 to 0.94. Feature-based Random Forest demonstrated the best performance among the machine learning models, with an accuracy of 0.88 for Mancos and 0.94 for Marcellus. The U-Net model with the linear combination of focal and dice loss also performed well with an accuracy of 0.91 and 0.93 for Mancos and Marcellus, respectively. In general, considering more features provided promising and reliable segmentation results that are valuable for analyzing the composition of dense samples, such as shales, which are significant unconventional reservoirs in oil recovery.